High-efficiency quantum photonics

Abstract

This thesis examines the integration of two disparate technologies in order to perform experiments in single photon quantum optics with low loss. Many technologies and experiments in quantum optics, communication, or computing require a certain fraction of the photons involved to be received at the end of the experiment, and in many cases the required efficiency has not yet been reached. This includes the famous Einstein, Podolsky, and Rosen gedanken-experiment, now implemented in the laboratory, demonstrating the existence of entanglement to unconvinced observers. Also included is the nonlocality test of John Bell, as well as technological problems such as device-independent quantum key distribution. In this thesis I perform these experiments in the high-efficiency regime. This programme requires the integration of two lines of research: improving sources of single photons, and improving detectors thereof. Until recently detector research was focussed on development, with improvements being sought for their own sake, working towards the ultimate goal of perfect photon detection. Recent years have seen these devices move into quantum photonics laboratories, allowing for previously impossible experiments to be undertaken. In this thesis, I combine high-efficiency, number-resolving, detectors with a high-efficiency entangled photon pair source, based on another line of research going back decades: the use of spontaneous parametric down-conversion to create single-photon-like modes of light, and the entanglement of the output modes in a useful way. For small demonstrations, such as the experiments mentioned above, this can emulate a single photon with enough fidelity, and low enough loss, to successfully perform the procedure. Here I push both of these technologies, and the problems of combining them, as far as I can, and solve the problems inherent with any marriage of disparate devices. I also examine the relative performance of two different steering inequalities, one linear and one quadratic, in the presence of noise, analysing my experimental data with each of them. I perform a detection- loophole free demonstration of EPR steering, violating the local bound by 200s, setting a then world record for efficiency at 62.5%.